Deallocated block determination

A first data block on a storage device including a data structure of deallocated data blocks on the storage device and a corresponding program erase count value for each of the deallocated data blocks is identified. A determination as to whether a second data block from the data structure of deallocated data blocks remains deallocated after being added to the data structure of deallocated data blocks based on the program erase count value is made. The data is stored at the second data block upon determining that the second data block remains deallocated after being added to the data structure of deallocated data blocks.

Skip to: Description  ·  Claims  ·  References Cited  · Patent History  ·  Patent History
Description
CROSS REFERENCE TO RELATED APPLICATIONS

This is a continuation application for patent entitled to a filing date and claiming the benefit of earlier-filed U.S. patent application Ser. No. 17/401,436, filed Aug. 13, 2021, which is a continuation of U.S. Pat. No. 11,119,657, issued Sep. 14, 2021, which is a continuation of U.S. Pat. No. 10,481,798, issued Nov. 19, 2019, each of which are hereby incorporated by reference in their entirety.

BACKGROUND

As computer memory storage and data bandwidth increase, so does the amount and complexity of data that businesses manage daily. Large-scale distributed storage systems, such as data centers, typically run many business operations. A datacenter, which also may be referred to as a server room, is a centralized repository, either physical or virtual, for the storage, management, and dissemination of data pertaining to one or more businesses. A distributed storage system may be coupled to client computers interconnected by one or more networks. If any portion of the distributed storage system has poor performance, company operations may be impaired. A distributed storage system therefore maintains high standards for data availability and high-performance functionality.

BRIEF DESCRIPTION OF THE DRAWINGS

The present disclosure is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings.

FIG. 1 is a block diagram illustrating a storage system in which embodiments of the present disclosure may be implemented.

FIG. 2 is a block diagram illustrating a memory manager in a storage controller, according to an embodiment.

FIG. 3 is a flow diagram illustrating a method for performing a metadata scan to populate a data structure, according to an embodiment.

FIG. 4 is a flow diagram illustrating a method for identifying a deallocated data block on a storage device, according to an embodiment.

FIG. 5 is a flow diagram illustrating a method for identifying leading edge data on a storage device, according to an embodiment.

FIG. 6 is a block diagram illustrating an exemplary computer system, according to an embodiment.

 DETAILED DESCRIPTION

Embodiments are described for efficient flash management for multiple controllers. In one embodiment, a memory manager module running on a storage controller utilizes physical block addressing rather than logical addressing to manage the data stored on the underlying storage devices in a connected storage array. The memory manager may abide by certain principles including having no preference for particular physical addresses, such that data does not have a “primary” location, “secondary” location, etc., but rather is just scattered randomly about the drive. Another principle of operations is not trying to write a “trickle” of tiny metadata updates because as the drive's state evolves, it may be best to use only the metadata persisted into flash blocks, along with periodically persisted data managed by the memory manager.

In one embodiment, the memory manager described herein achieves these principles by meeting at least three individual objectives. First, the memory manager allows a fast start of the primary storage controller by quickly locating recently-written flash blocks of one or more types of data. Second, the memory manager provides a list of flash blocks that are deallocated and ready for client writes. Third, the memory manager respects the needs of modern flash devices by delaying the erase of flash blocks until just before those blocks are needed for writing.

To accomplish the above objectives, the memory manager works in connection with firmware in the underlying storage devices. In one embodiment, the storage device firmware implements a metadata tracking scheme that stores certain values along with any data payload. These values may include a program/erase count for each data block that indicates a number of cycles during which the block has been written and erased, and a block type value, which may identify a storage client that owns the data in that block. In addition, the storage device firmware maintains a table or other data structure containing data for each data block and allows the memory manager to access the table. For each data block, the table also stores the program/erase count, the block type value, and a block status indicator (erased, written, unreadable, bad, etc.). On power-up of the storage device, the firmware may scan the data blocks to recover the embedded metadata and populate the data structure with the metadata recovered from those blocks.

The memory manager manages blocks in concert with a data structure called the “frontier set.” The frontier set is a data structure that is written to flash that declares the state of a storage device in a way that allows future readers to determine not only what was true at that point in time, but also to recover the effect of operations that occurred after the frontier set was written to flash. The frontier set, in its most primitive form, is simply a list of block numbers and their corresponding program/erase counts. This is a declaration that at the moment the frontier set was created, block X was deallocated when its program/erase count was Y. The memory manager can use the frontier set to extrapolate the state of a block beyond the moment the frontier set was created and written. If block X was deallocated when its program/erase count was Y, the memory manager can make at least two logical conclusions. First, a future reader that finds block X still at program/erase count Y can conclude that block must still be deallocated. Second, a future reader that finds block X at some program/erase count Z>Y can conclude that some client must have written new data to that block, after this frontier set was created or updated.

As explained herein, these conclusions allow the memory manager to achieve the objectives described above. As long as the memory manager writes a new frontier set periodically (and sufficient blocks remain available), it is possible to allow the controller to discover known deallocated blocks that are ready for new writes. In addition, erases can be delayed until just before the blocks are rewritten because block deallocation will be “eventually consistent.” This means that a deallocated block may not be seen as deallocated by all possible future primary controllers since it's possible that deallocated blocks may revert to the allocated state, until the next frontier set is persisted. Deallocated blocks are available for reuse, but until the moment that their new owner actually writes to them (which implies that the embedded metadata will include a new program/erase count and block type value), that block may revert back to its previous owner. Furthermore, a fast start of the primary controller is achieved by locating the leading edge of newly written data from a particular client. To assist a major client storing on the order of 100,000 to 1,000,000 blocks or more, the memory manager can define two classes of data, or more precisely, two states that a data block can be in. A “boot” block is one that contains new data and a “standalone” block contains cold data (i.e., data that has been around and untouched for a certain period of time). Boot blocks can be quickly and efficiently enumerated by memory manager to the client, after a crash or power loss. When the client no longer requires this block to be enumerated as a boot block, it will indicate this to the memory manager (a process referred to as “graduation”). This block will then become a standalone block at the memory manager's discretion.

FIG. 1 is a block diagram illustrating a storage system 100 in which embodiments of the present disclosure may be implemented. Storage system 100 may include storage controllers 110, 150 and storage array 130, which is representative of any number of data storage arrays or storage device groups. As shown, storage array 130 includes storage devices 135A-n, which are representative of any number and type of storage devices (e.g., solid-state drives (SSDs)). Storage controller 110 may be coupled directly to initiator device 125 and storage controller 110 may be coupled remotely over network 120 to initiator device 115. In one embodiment, storage controller 150 is coupled remotely over network 120 to initiator device 115. Initiator devices 115 and 125 are representative of any number of clients which may utilize storage controllers 110 and 150 for storing and accessing data in storage system 100. It is noted that some systems may include only a single client or initiator device, connected directly or remotely, to storage controllers 110 and 150.

In one embodiment, controller 110 is designated as the “primary” controller, which performs most or all of the I/O operations on the array 130. If, however, a software crash, hardware fault or other error occurs, the “secondary” controller 150 may be promoted to serve as the primary controller and take over all responsibilities for servicing the array 130. In one embodiment, storage controllers 110 and 150 are identical and any description of controller 110 herein may be equally attributed to storage controller 150.

Storage controller 110 may include software and/or hardware configured to provide access to storage devices 135A-n. Although storage controller 110 is shown as being separate from storage array 130, in some embodiments, storage controller 110 may be located within storage array 130. Storage controller 110 may include or be coupled to a base operating system (OS), a volume manager, and additional control logic, such as memory manager 140, for implementing the various techniques disclosed herein. In one embodiment, the OS is designed with flash storage in mind, so while it can use conventional SSDs to store data, it does not depend on a 512 byte random overwrite capability. Even on conventional SSDs, storage controller 110 can achieve better performance by writing and discarding data in large chunks. This style of I/O is sometimes called “flash friendly I/O.” This also makes it a much easier task to convert the OS to use the physical addressing of storage devices, as compared to conventional filesystems.

In one embodiment, the logic of memory manager 140 is contained within an object which manages one of devices 135A-n. Thus, there may be a separate memory manager object for each device 135A-n in storage array 130. As new devices are connected to controller 110, new memory manager objects may be created. These objects may be similarly discarded when a corresponding device is disconnected from storage controller 110. Clients wishing to communicate with memory manager 140, such as one of initiator applications 112, 122, the operating system running on storage controller 110 or another client application running on storage controller 110, may do so via a memory manager application programming interface (API) published by memory manager 140. In one embodiment, multiple clients can access the same memory manager object concurrently. In one embodiment, storage controller 150 includes a separate instance(s) of memory manager 152.

Storage controller 110 may include and/or execute on any number of processing devices and may include and/or execute on a single host computing device or be spread across multiple host computing devices, depending on the embodiment. In some embodiments, storage controller 110 may generally include or execute on one or more file servers and/or block servers. Storage controller 110 may use any of various techniques for replicating data across devices 135A-n to prevent loss of data due to the failure of a device or the failure of storage locations within a device. Storage controller 110 may also utilize any of various deduplication techniques for reducing the amount of data stored in devices 135A-n by deduplicating common data.

In one embodiment, storage controller 110 may utilize logical volumes and mediums to track client data that is stored in storage array 130. A medium is defined as a logical grouping of data, and each medium has an identifier with which to identify the logical grouping of data. A volume is a single accessible storage area with a single file system, typically, though not necessarily, resident on a single partition of a storage device. The volumes may be logical organizations of data physically located on one or more of storage device 135A-n in storage array 130. Storage controller 110 may maintain a volume to medium mapping table to map each volume to a single medium, and this medium is referred to as the volume's anchor medium. A given request received by storage controller 110 may indicate at least a volume and block address or file name, and storage controller 110 may determine an anchor medium targeted by the given request from the volume to medium mapping table.

In one embodiment, storage controller 110 includes memory manager 140. Memory manager 140 can perform various operations to identify deallocated data blocks available for writing and to identify leading edge data that was most recently written by a particular client. In one embodiment, memory manager 140 can receive a request to write data to a storage device 135A and can determine a first data block on storage device 135A comprising a list of deallocated data blocks. That list may include a block number of each deallocated data block and an access operation count value (e.g., program/erase count value) at which each deallocated data block was deallocated. Memory manager 140 can then identify a second data block from the list of deallocated data blocks and write the requested data to that second data block. To identify the leading edge data, memory manager 140 may access a data structure stored in memory on storage device 135A, where the data structure stores block metadata for each data block on storage device 135A. Memory manager 140 may determine, from the data structure, a first data block on storage device 135A comprising a list of deallocated data blocks on the storage device and compare a first access operation count value associated with each of the deallocated data blocks from the data structure to a second access operation count value associated with each of the deallocated data blocks from the list of deallocated data blocks. Memory manager 140 may label a second data block on the list as comprising new data responsive to the first access operation count value associated with the second data block from the data structure not matching the second access operation count value associated with the second data block from the list of deallocated data blocks.

In various embodiments, multiple mapping tables may be maintained by storage controller 110. These mapping tables may include a medium mapping table and a volume to medium mapping table. These tables may be utilized to record and maintain the mappings between mediums and underlying mediums and the mappings between volumes and mediums. Storage controller 110 may also include an address translation table with a plurality of entries, wherein each entry holds a virtual-to-physical mapping for a corresponding data component. This mapping table may be used to map logical read/write requests from each of the initiator devices 115 and 125 to physical locations in storage devices 135A-n. A “physical” pointer value may be read from the mappings associated with a given medium during a lookup operation corresponding to a received read/write request. The term “mappings” is defined as the one or more entries of the address translation mapping table which convert a given medium ID and block number into a physical pointer value. This physical pointer value may then be used to locate a physical location within the storage devices 135A-n. The physical pointer value may be used to access another mapping table within a given storage device of the storage devices 135A-n. Consequently, one or more levels of indirection may exist between the physical pointer value and a target storage location.

In alternative embodiments, the number and type of client computers, initiator devices, storage controllers, networks, storage arrays, and data storage devices is not limited to those shown in FIG. 1. At various times one or more clients may operate offline. In addition, during operation, individual client computer connection types may change as users connect, disconnect, and reconnect to storage system 100. Further, the systems and methods described herein may be applied to directly attached storage systems or network attached storage systems and may include a host operating system configured to perform one or more aspects of the described methods. Numerous such alternatives are possible and are contemplated.

Network 120 may utilize a variety of techniques including wireless connections, direct local area network (LAN) connections, wide area network (WAN) connections such as the Internet, a router, storage area network, Ethernet, and others. Network 120 may comprise one or more LANs that may also be wireless. Network 120 may further include remote direct memory access (RDMA) hardware and/or software, transmission control protocol/internet protocol (TCP/IP) hardware and/or software, router, repeaters, switches, grids, and/or others. Protocols such as Fibre Channel, Fibre Channel over Ethernet (FCoE), iSCSI, and so forth may be used in network 120. The network 120 may interface with a set of communications protocols used for the Internet such as the Transmission Control Protocol (TCP) and the Internet Protocol (IP), or TCP/IP. In one embodiment, network 120 represents a storage area network (SAN) which provides access to consolidated, block level data storage. The SAN may be used to enhance the storage devices accessible to initiator devices so that the devices 135A-n appear to the initiator devices 115 and 125 as locally attached storage.

Initiator devices 115 and 125 are representative of any number of stationary or mobile computers such as desktop personal computers (PCs), servers, server farms, workstations, laptops, handheld computers, servers, personal digital assistants (PDAs), smart phones, and so forth. Generally speaking, initiator devices 115 and 125 include one or more processing devices, each comprising one or more processor cores. Each processor core includes circuitry for executing instructions according to a predefined general-purpose instruction set. For example, the x86 instruction set architecture may be selected. Alternatively, the ARM®, Alpha®, PowerPC®, SPARC®, or any other general-purpose instruction set architecture may be selected. The processor cores may access cache memory subsystems for data and computer program instructions. The cache subsystems may be coupled to a memory hierarchy comprising random access memory (RAM) and a storage device.

In one embodiment, initiator device 115 includes initiator application 112 and initiator device 125 includes initiator application 122. Initiator applications 112 and 122 may be any computer application programs designed to utilize the data on devices 135A-n in storage array 130 to implement or provide various functionalities. Initiator applications 112 and 122 may issue requests to read or write data from certain logical volumes data within storage system 100. Those requests can be serviced by memory manager 140 of storage controller 110, as described in detail herein.

FIG. 2 is a block diagram illustrating memory manager 140 in a storage controller 110, according to an embodiment. In one embodiment, memory manager 140 includes client interface 242, data structure interface 244, data block interface 246 and comparison logic 248. This arrangement of modules may be a logical separation, and in other embodiments, these modules, interfaces or other components can be combined together or separated in further components. In one embodiment, storage device 135A is connected to memory manager 140 and includes firmware 252, memory 235 storing data structure 254, and data blocks 256. In one embodiment, storage device 135A may be external to storage controller 110 as part of storage array 130 and may be connected to storage controller 110 over a network or other connection. In other embodiments, storage controller 110 may include different and/or additional components which are not shown to simplify the description. Storage device 135A may include one or more mass storage devices which can include, for example, flash memory or solid-state drives (SSDs). Memory 235 may include for example, random-access memory (RAM); dynamic random-access memory (DRAM); erasable programmable memory (e.g., EPROM and EEPROM); flash memory; or any other type of storage medium. In one embodiment, storage device 135A includes volatile memory 235, such as DRAM, and non-volatile data blocks 256, such as flash blocks or other persistent data blocks.

In one embodiment, client interface 242 manages communication with client devices or applications in storage system 100, such as initiator devices 115 or 125, or applications within storage controller 110. Client interface 242 can receive I/O requests to access data blocks 256 on storage device 135A from an initiator application 112 or 122 over network 120. In one embodiment, the I/O request includes a request to write new data to storage device 135A. After the write is performed, client interface may provide a notification to initiator device 115 or 125 over network 120 indicating that the write was successfully performed.

In one embodiment, data structure interface 244 interacts with data structure 254 in memory 235 on storage device 135A. In response to client interface 242 receiving a write request, for example, data structure interface 244 may access data structure 254 (e.g., a dynamic table) comprising block metadata for each of data blocks 256 on storage device 135A. The block metadata may include an indication of a block type of each data block 256 and an access operation count value for each data block 256. In one embodiment, the access operation count value is a total number of program/erase cycles that have been performed on the block. Using the block type indicator, data structure interface 244 may determine a first data block on storage device 135A which stores a list of deallocated data blocks on storage device 135A. This list may include a block number of each deallocated data block and an access operation count value at which each deallocated data block was deallocated.

In one embodiment, data block interface 246 interacts with data blocks 256 of storage device 135A as part of any data access operations being performed. For example, once data structure interface 244 determines the block storing the list of deallocated data blocks, data block interface 246 may identify a second block of those deallocated blocks from the list, and read an access operation count value associated with the second block from the list. If memory manager 140 ultimately determines that the second block was in fact deallocated, data block interface 246 may perform the requested write operation by overwriting the old data in the second data block with new data. If the allegedly deallocated data block was not actually deallocated (or has since been reallocated), data block interface 246 can remove the second data block from the list stored in the first data block. When memory manager 140 is attempting to locate leading edge data, data block interface 246 can determine whether a particular block was previously labeled as comprising new data. In addition, once memory manager 140 identifies the leading edge data, data block interface 246 can label the data blocks as comprising either new or old data, as appropriate.

In one embodiment, comparison logic 248 performs various calculations and comparisons as part of the operations performed by memory manager 140. For example, to verify that a block appearing on the list of deallocated data blocks is in fact deallocated or to determine whether a block is storing new or old data, comparison logic 248 may compare a first access operation count value associated with the data block from data structure 254 to a second access operation count value associated with the data block from the list of deallocated data blocks stored in one of data blocks 256 (identified by the block type value). If comparison logic 248 determines that the count values match, this indicates that the block has not been reallocated since it was added to the list of deallocated blocks and, thus, can either be labeled as storing old data or can safely be overwritten without losing any critical data. If the count values don't match, however, this indicates that another client has written data to that block and it should be removed from the list of deallocated blocks and/or labeled as comprising new data.

FIG. 3 is a flow diagram illustrating a method for performing a metadata scan to populate a data structure, according to an embodiment. The method 300 may be performed by processing logic that comprises hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, etc.), software (e.g., instructions run on a processing device to perform hardware simulation), firmware, or a combination thereof. On power-up of the storage device 135A, the method 300 may scan the data blocks 256 to recover the embedded metadata and populate the data structure 254 with the metadata recovered from those blocks. In one embodiment, method 300 may be performed by the firmware 252 of storage device 135A, as shown in FIG. 2.

Referring to FIG. 3, at block 310, method 300 detects a restart of storage device 135A. After an event, such as a sudden power loss or software crash, storage device 135A may be automatically restarted. Firmware 252 can detect the restart and initiate any number of start-up procedures. Storage system 100 is designed to be extremely reliable even in the face of hardware or software failures, including sudden power losses or software crashes. Storage devices 135A-n may be designed with failures in mind, specifically to account for sudden crashes or power loss at any point during the process, and to assure that restarting after such an event is no different than any other restart operation.

At block 320, method 300 scans a plurality of data blocks 256 on storage device 135A to identify block metadata for each of the plurality of data blocks 256. In one embodiment, firmware 252 may scan each of the data blocks 256 to recover the embedded metadata. This metadata may include, for example, an indication of a block type of each of the data blocks 256 and an access operation count value for each of the data blocks 256. In one embodiment, this metadata may be stored in a header section of each individual block 256, so that it can be obtained quickly and efficiently. In one embodiment, the scan is reasonably fast (e.g., taking less than 10 seconds to scan one million or more blocks).

At block 330, method 300 stores the block metadata in data structure 254 stored in memory 235 on storage device 135A, wherein the block metadata is accessible by storage controller 110 coupled to storage device 135A. In one embodiment, data structure 254 comprises a plurality of entries, each of the entries corresponding to a different one of the data blocks 256 on storage device 135A. In one embodiment, data structure 254 maintains an indication of a block type and an access operation count value for each data block. One example of a block type that may be stored in metadata is a “bootstrap data” block. This block type stores data that is needed to restart the system after a power loss or other failure. Since the data blocks 256 may have limited reusability (e.g., approximately three thousand program/erase cycles), this data cannot be stored in the same place since it gets accessed regularly. Since the bootstrap data can be located easily during the metadata scan, it can be located anywhere on storage device 135A.

Data storage on flash follows a simple looking cycle: erase, program; erase, program. Once written or programmed, data is nonvolatile and may be read millions of times. Flash structures have a complex hierarchy (i.e., packages, dies, planes, blocks, pages, and bits). The memory manager 140 described herein operates primarily on the block level. This is because the block is the most common unit of erase and reuse. If a particular flash chip uses 16 MB blocks, then data will be written to flash in 16 MB chunks, and will be discarded in 16 MB chunks. One physical detail of NAND flash that becomes important is that the physical blocks and pages have some “extra” storage beyond the expected powers of two. Thus, a physical 16 KB block may actually contain 19 KB of physical bits. While most of these “extra” bits are consumed by error correction codes, there may be some room left over to store metadata about the block or its contents. Storage system 100 makes use of some of these bits to store metadata alongside any data stored by the controller.

There are a number of management tasks performed by any system that stores data on NAND flash chips. Normally these functions are all performed in SSD firmware and are concealed from the host computer. Flash chips have a limited lifespan, measured in program/erase (PE) cycles. Management software must spread data around such that flash blocks wear more or less evenly across the drive, or premature drive failure may result due to block failures. This may be referred to as “wear leveling.” Flash is an imperfect media and blocks may fail spontaneously. Thus, in one embodiment, management software must maintain a bad block list over the lifetime of the drive. In addition, most SSDs support a storage interface that is backwards compatible with hard disks from the last 20+ years, allowing a contiguous range of logical 512 byte sectors that can be overwritten randomly an arbitrary number of times. Firmware supports this interface via a complicated abstraction layer that maps logical addresses to physical flash locations dynamically to provide logical address mapping. In one embodiment, wear leveling and bad block handling are not performed in firmware, but rather are handled within memory manager 140 of storage controller 110. Thus, the logical address feature may be discarded as only physical flash addresses are used.

FIG. 4 is a flow diagram illustrating a method for identifying a deallocated data block on a storage device, according to an embodiment. The method 400 may be performed by processing logic that comprises hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, etc.), software (e.g., instructions run on a processing device to perform hardware simulation), firmware, or a combination thereof. The method 400 can allow a storage controller to identify deallocated data blocks to allow those blocks to be overwritten without the risk of losing any critical data. In one embodiment, method 400 may be performed by memory manager 140, as shown in FIGS. 1 and 2.

Referring to FIG. 4, at block 410, method 400 receives a request to write data to storage device 135A (e.g., an SSD). In one embodiment, client interface 242 receives the request from an initiator application 112 or 122 over network 120, or from another client application on storage controller 110. The write request may not specify a particular data block on storage device 135A, so it may be up to memory manager 140 to identify a deallocated block. Since data blocks 256 may not be erased until right before they are written, even a block currently storing data may be “deallocated” and, thus, available for writing.

At block 420, method 400 accesses a dynamic table (e.g., data structure 254) stored in memory 235 on storage device 135A, the dynamic table comprising block metadata for each data block 256 on storage device 135A. In one embodiment, data structure interface 244 may access data structure 254 comprising block metadata for each of data blocks 256 on storage device 135A. The block metadata may include an indication of a block type of each data block and an access operation count value for each data block.

At block 430, method 400 determines a first data block on storage device 135A comprising a list of deallocated data blocks on the storage device 135A. In one embodiment, data structure interface 244 determines the first data block from the dynamic table based on the indication of the block type of the first data block. Using the block type indicator stored in the dynamic table, data structure interface 244 may determine a first data block on storage device 135A which stores a list of deallocated data blocks on storage device 135A. This list may include a block number of each deallocated data block and an access operation count value at which each deallocated data block was deallocated. In one embodiment, this list may be referred to as the “frontier set” and the block where it is stored may be given a special block type. In one embodiment, the frontier set is identified and read once when storage device 135A is started-up (or restarted), and the list of deallocated blocks is stored and their corresponding operation count values are stored in memory 235.

At block 440, method 400 identifies a second data block from the list of deallocated data blocks on the storage device 135A. In one embodiment, data block interface 246 accesses the first block on storage device 135A identified at block 430 from the dynamic table. Data block interface 246 may identify a second block of those deallocated blocks from the list, and read an access operation count value associated with the second block from the list. In one embodiment, the second data block may be identified from the list of deallocated blocks stored in memory 235 some period of time after the frontier set is initially identified at block 430.

At block 450, method 400 compares a first access operation count value associated with the second data block from the dynamic table to a second access operation count value associated with the second data block from the list of deallocated data blocks. In one embodiment, comparison logic 248 compares the first access operation count value to the second access operation count value associated with the data block. If comparison logic 248 determines that the count values match, this indicates that the block has not been reallocated since it was added to the list of deallocated blocks. Thus, at block 460, method 400 writes the requested data to the second data block. In one embodiment, data block interface 246 overwrites the old data stored in the second data block with the newly requested data received at block 410. If the count values do not match, however, this indicates that another client has since written new data to that block. Thus, at block 470, data block interface 246 removes the second data block from the list of deallocated data blocks.

FIG. 5 is a flow diagram illustrating a method for identifying leading edge data on a storage device, according to an embodiment. The method 500 may be performed by processing logic that comprises hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, etc.), software (e.g., instructions run on a processing device to perform hardware simulation), firmware, or a combination thereof. The method 500 can allow a storage controller to identify leading edge data that was most recently written to storage device 135A by a particular storage client. In one embodiment, method 500 may be performed by memory manager 140, as shown in FIGS. 1 and 2.

Referring to FIG. 5, at block 510, method 500 accesses data structure 254 in memory 235 on storage device 135A, the data structure 254 comprising block metadata for each data block 256 on storage device 135A. In one embodiment, data structure interface 244 may access data structure 254 comprising block metadata for each of data blocks 256 on storage device 135A. The block metadata may include an indication of a block type of each data block and an access operation count value for each data block.

At block 520, method 500 determines, from data structure 254, a first data block on storage device 135A comprising a list of deallocated data blocks on storage device 135A. In one embodiment, data structure interface 244 determines the first data block from the data structure 254 based on the indication of the block type of the first data block. Using the block type indicator stored in the data structure 254, data structure interface 244 may determine a first data block on storage device 135A which stores a list of deallocated data blocks on storage device 135A. This list may include a block number of each deallocated data block and an access operation count value at which each deallocated data block was deallocated and a state value indicating whether the block is known to be in use. The state value may be indicated as “boot” or “future,” where a boot block is known to be in use (because it includes new data) and a future block was, at the time data structure 254 was populated, not in use (because it includes old data). In one embodiment, this list may be referred to as the “frontier set” and the block where it is stored may be given a special block type. The frontier set represents a snapshot in time and permits a future primary controller to correctly recover the state of the drive (and all the clients' data storage). A newly started primary controller can examine this frontier set, and compare the program/erase counts to those on the drive itself. “Future” blocks will become boot blocks if the program/erase count indicates new writes have occurred. “Boot” blocks will generally stay boot blocks, barring unusual events, such as flash errors, etc.

At block 530, method 500 determines whether a data block on the list of deallocated data blocks was previously labeled as comprising old data. In one embodiment, data block interface 246 reads the data from the first data block where the frontier set itself is stored, locates an entry in the frontier set corresponding to a second data block and reads the state value for that entry. As described above, data blocks on storage device 135A comprising old data may be labeled as “future” blocks in the frontier set and data blocks comprising new data may be labeled as “boot” blocks in the frontier set.

If the block was previously labeled as comprising new data (i.e., labeled as “boot” blocks), at block 540, method 500 maintains the previous new data label. Thus, boot blocks stay labeled boot blocks, regardless of whether they have new writes.

If the block was previously labeled as comprising old data, at block 550, method 500 compares a first access operation count value associated with the deallocated data blocks from the data structure 254 to a second access operation count value associated with the deallocated data block from the list of deallocated data blocks. In one embodiment, comparison logic 248 compares the first access operation count value to the second access operation count value associated with the data block. If comparison logic 248 determines that the first access operation count value associated with the second data block from the data structure does not match the second access operation count value associated with the second data block from the list of deallocated data blocks, this indicates that the block has been written with new data since it was added to the list. Thus, at block 540, method 500 labels the second data block on the list as comprising new data. In one embodiment, data block interface 246 changes the state value in the entry of the frontier set corresponding to the data block to “boot.” If the first access operation count value associated with the second data block from the data structure matches the second access operation count value associated with the second data block from the list of deallocated data blocks, however, this indicates that no new data has been written to that block. Thus, at block 560, method 500 labels the second data block on the list as comprising old data. In one embodiment, data block interface 246 maintains the state value in the entry of the frontier set corresponding to the data block as “future.”

In one embodiment, standalone blocks are not tracked by the frontier set. Rather, they can be defined as “all the blocks not listed in the frontier set” (or marked bad). These blocks can be enumerated simply by searching the drive tables for blocks with the correct block type that are not covered by the current frontier set. It should also be noted that all newly allocated blocks may come from the list of blocks written in the frontier set. Otherwise, the newly written data would not be detected as a boot block by a future primary controller. Furthermore, because boot blocks are listed in the frontier set, there can be policy on clients with small storage needs that they will never use standalone blocks. This means that any blocks that contain data with that client's block type, but are not listed in the frontier set, have been deallocated, saving clients the bother of sifting through old data and deallocating those blocks.

Periodically, a new frontier set may be constructed and written to flash. This may be desirable if most of the deallocated blocks in the current frontier set have been exhausted. Writing a new frontier set can make additional blocks ready for allocation. In addition, if a particular client has written and then deallocated many blocks, a new frontier set may be created. In one embodiment, deallocations are not made permanent until a new frontier set is written. Thus, completing that write will limit the number of blocks a future primary will discover. This can affect failover time, so it may be advantageous to limit some clients' block usage. Failover includes the transfer of the “primary” designation from one controller to another, which includes the transfer of the functionality of memory manager 140. Furthermore, when some blocks with attractive (i.e., low) program/erase counts have become deallocated, for best wear leveling behavior, the system tries to use the blocks with the lowest program/erase count first. Persisting a new frontier set can make those more attractive blocks available.

A new frontier set may be constructed by selecting some new future blocks from the list of deallocated blocks and adding the list of known boot blocks. Blocks which were formerly boot blocks in the previous frontier set but have since been graduated may not be recorded in the new frontier set. While persisting a new frontier set, the system need not halt all client allocations and writes. The new frontier set is constructed to contain at least a small number of blocks that are also part of the previous frontier set (and are marked as “future” blocks). In this way the system can continue allocating blocks from this overlapping set, knowing that if it crashes or loses power at any time, memory manager 140 can safely start up using either the new set (assuming the new frontier set write completed) or the old set (if the write did not complete). Each new frontier set may be written to a flash block using a “frontier set” block type value. This assists in quickly locating the frontier set.

Because this design does not persist a new frontier set during every state change, there may be an accumulation of changes that have only taken effect inside the controller memory. These changes can be lost if a failover or power loss occurs. It is possible to lose block deallocations and graduations across these events, however, this does not affect the correctness of the controller application, as it is straightforward to simply repeat these actions on the new primary controller. It is also possible to lose a block allocation if the client did not write to the block in question before the failover or restart. In one embodiment, this problem is alleviated as whatever task that intended to write something will be repeated on the new primary controller and may allocate a new block without being aware that a previous primary controller was attempting a similar operation.

In one embodiment, the storage device firmware 252 provides a list of blocks containing data of a particular block type to memory manager 140. This reduces the work required to locate the frontier set, and allows faster primary controller startup. When writing the frontier set, memory manager 140 also takes the unusual step of physically erasing old copies of frontier sets. This allows faster startup time by ensuring that controller 110 does not have to sift through many old frontier sets looking for the current one. To ensure correct behavior across a surprise block failure, frontier sets may be written to at least two physical locations, preferably spread across different failure domains (e.g. flash dies) to minimize the chance that both fail simultaneously. In one embodiment, the encryption capabilities of storage device firmware 252 are customized such that the data structures needed during startup (e.g. the frontier set) can be written unencrypted, and most client data can be written encrypted.

FIG. 6 illustrates a diagrammatic representation of a machine in the exemplary form of a computer system 600 within which a set of instructions, for causing the machine to perform any one or more of the methodologies discussed herein, may be executed. In alternative embodiments, the machine may be connected (e.g., networked) to other machines in a local area network (LAN), an intranet, an extranet, or the Internet. The machine may operate in the capacity of a server or a client machine in a client-server network environment, or as a peer machine in a peer-to-peer (or distributed) network environment. The machine may be a personal computer (PC), a tablet PC, a set-top box (STB), a Personal Digital Assistant (PDA), a cellular telephone, a web appliance, a server, a network router, switch or bridge, or any machine capable of executing a set of instructions (sequential or otherwise) that specify actions to be taken by that machine. Further, while only a single machine is illustrated, the term “machine” shall also be taken to include any collection of machines that individually or jointly execute a set (or multiple sets) of instructions to perform any one or more of the methodologies discussed herein. In one embodiment, computer system 600 may be representative of a server, such as storage controller 110 running memory manager 140 or of a client, such as initiator devices 115 or 125.

The exemplary computer system 600 includes a processing device 602, a main memory 604 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM), a static memory 606 (e.g., flash memory, static random access memory (SRAM), etc.), and a data storage device 618, which communicate with each other via a bus 630. Data storage device 618 may be one example of any of the storage devices 135A-n in FIGS. 1 and 2. Any of the signals provided over various buses described herein may be time multiplexed with other signals and provided over one or more common buses. Additionally, the interconnection between circuit components or blocks may be shown as buses or as single signal lines. Each of the buses may alternatively be one or more single signal lines and each of the single signal lines may alternatively be buses.

Processing device 602 represents one or more general-purpose processing devices such as a microprocessor, central processing unit, or the like. More particularly, the processing device may be complex instruction set computing (CISC) microprocessor, reduced instruction set computer (RISC) microprocessor, very long instruction word (VLIW) microprocessor, or processor implementing other instruction sets, or processors implementing a combination of instruction sets. Processing device 602 may also be one or more special-purpose processing devices such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), network processor, or the like. The processing device 602 is configured to execute processing logic 626, which may be one example of memory manager 140 shown in FIGS. 1 and 2, or of initiator application 112 or 122, for performing the operations and steps discussed herein.

The data storage device 618 may include a machine-readable storage medium 628, on which is stored one or more set of instructions 622 (e.g., software) embodying any one or more of the methodologies of functions described herein, including instructions to cause the processing device 602 to execute virtual copy logic 140 or initiator application 112 or 122. The instructions 622 may also reside, completely or at least partially, within the main memory 604 and/or within the processing device 602 during execution thereof by the computer system 600; the main memory 604 and the processing device 602 also constituting machine-readable storage media. The instructions 622 may further be transmitted or received over a network 620 via the network interface device 608.

The machine-readable storage medium 628 may also be used to store instructions to perform a method for efficient flash management for multiple controllers, as described herein. While the machine-readable storage medium 628 is shown in an exemplary embodiment to be a single medium, the term “machine-readable storage medium” should be taken to include a single medium or multiple media (e.g., a centralized or distributed database, and/or associated caches and servers) that store the one or more sets of instructions. A machine-readable medium includes any mechanism for storing information in a form (e.g., software, processing application) readable by a machine (e.g., a computer). The machine-readable medium may include, but is not limited to, magnetic storage medium (e.g., floppy diskette); optical storage medium (e.g., CD-ROM); magneto-optical storage medium; read-only memory (ROM); random-access memory (RAM); erasable programmable memory (e.g., EPROM and EEPROM); flash memory; or another type of medium suitable for storing electronic instructions.

The preceding description sets forth numerous specific details such as examples of specific systems, components, methods, and so forth, in order to provide a good understanding of several embodiments of the present disclosure. It will be apparent to one skilled in the art, however, that at least some embodiments of the present disclosure may be practiced without these specific details. In other instances, well-known components or methods are not described in detail or are presented in simple block diagram format in order to avoid unnecessarily obscuring the present disclosure. Thus, the specific details set forth are merely exemplary. Particular embodiments may vary from these exemplary details and still be contemplated to be within the scope of the present disclosure.

In situations in which the systems discussed herein collect personal information about users, or may make use of personal information, the users may be provided with an opportunity to control whether programs or features collect user information (e.g., information about a user's social network, social actions or activities, profession, a user's preferences, or a user's current location), or to control whether and/or how to receive content from the media server that may be more relevant to the user. In addition, certain data may be treated in one or more ways before it is stored or used, so that personally identifiable information is removed. For example, a user's identity may be treated so that no personally identifiable information can be determined for the user, or a user's geographic location may be generalized where location information is obtained (such as to a city, ZIP code, or state level), so that a particular location of a user cannot be determined. Thus, the user may have control over how information is collected about the user and used by the web server or media server.

Reference throughout this specification to “one embodiment” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiments included in at least one embodiment. Thus, the appearances of the phrase “in one embodiment” or “in an embodiment” in various places throughout this specification are not necessarily all referring to the same embodiment. In addition, the term “or” is intended to mean an inclusive “or” rather than an exclusive “or.”

Although the operations of the methods herein are shown and described in a particular order, the order of the operations of each method may be altered so that certain operations may be performed in an inverse order or so that certain operation may be performed, at least in part, concurrently with other operations. In another embodiment, instructions or sub-operations of distinct operations may be in an intermittent and/or alternating manner.

Claims

1. A system comprising:

a plurality of storage devices; and
a storage controller coupled to the plurality of storage devices, the storage controller comprising a processing device, the processing device configured to: identify a data structure containing metadata associated with a data block of the plurality of storage devices; identify a list of deallocated data blocks for the plurality of storage devices; compare an access operation count value for the data block associated with the data structure with an access operation count value for the data block associated with the list of deallocated blocks; and determine whether the data block is available for writing additional data based on comparison of access operation count values.

2. The system of claim 1, wherein identification of the data structure containing metadata is responsive to a restart of the system.

3. The system of claim 1, wherein the data structure containing metadata is a dynamic table comprising block metadata for each data block on the storage device.

4. The system of claim 3, wherein the block metadata comprises an indication of a block type of each data block on the storage device and a corresponding program erase count value for each data block on the storage device.

5. The system of claim 4, wherein the data block is identified from the dynamic table based on the indication of the block type of the data block.

6. The system of claim 1, wherein the data structure containing metadata is a frontier set.

7. The system of claim 1, wherein the processing device is configured to remove the data block from the list of deallocated data blocks in response to the access operation counts not matching.

8. A method comprising:

identifying a data structure containing metadata associated with a data block of the plurality of storage devices;
identifying a list of deallocated data blocks for the plurality of storage devices;
comparing an access operation count value for the data block associated with the data structure with an access operation count value for the data block associated with the list of deallocated blocks; and
determining whether the data block is available for writing additional data based on comparison of access operation count values.

9. The method of claim 8, wherein identification of the data structure containing metadata is responsive to a restart of the system.

10. The method of claim 8, wherein the data structure containing metadata is a dynamic table comprising block metadata for each data block on the storage device.

11. The method of claim 10, wherein the block metadata comprises an indication of a block type of each data block on the storage device and a corresponding program erase count value for each data block on the storage device.

12. The method of claim 11, wherein the data block is identified from the dynamic table based on the indication of the block type of the data block.

13. The method of claim 8, wherein the data structure containing metadata is a frontier set.

14. The method of claim 8, wherein the processing device is configured to remove the data block from the list of deallocated data blocks in response to the access operation counts not matching.

15. A non-transitory computer readable storage medium comprising instructions which, when executed by a processing device, cause the processing device to:

identify a data structure containing metadata associated with a data block of the plurality of storage devices;
identify a list of deallocated data blocks for the plurality of storage devices;
compare an access operation count value for the data block associated with the data structure with an access operation count value for the data block associated with the list of deallocated blocks; and
determine whether the data block is available for writing additional data based on comparison of access operation count values.

16. The non-transitory computer readable storage medium of claim 15, wherein identification of the data structure containing metadata is responsive to a restart of the system.

17. The non-transitory computer readable storage medium of claim 15, wherein the data structure containing metadata is a dynamic table comprising block metadata for each data block on the storage device.

18. The non-transitory computer readable storage medium of claim 17, wherein the block metadata comprises an indication of a block type of each data block on the storage device and a corresponding program erase count value for each data block on the storage device.

19. The non-transitory computer readable storage medium of claim 18, wherein the data block is identified from the dynamic table based on the indication of the block type of the data block.

20. The non-transitory computer readable storage medium of claim 15, wherein the data structure containing metadata is a frontier set.

Referenced Cited
U.S. Patent Documents
5390327 February 14, 1995 Lubbers et al.
5450581 September 12, 1995 Bergen et al.
5479653 December 26, 1995 Jones
5488731 January 30, 1996 Mendelsohn
5504858 April 2, 1996 Ellis et al.
5564113 October 8, 1996 Bergen et al.
5574882 November 12, 1996 Menon et al.
5649093 July 15, 1997 Hanko et al.
5883909 March 16, 1999 DeKoning et al.
6000010 December 7, 1999 Legg
6260156 July 10, 2001 Garvin et al.
6269453 July 31, 2001 Krantz
6275898 August 14, 2001 DeKoning
6453428 September 17, 2002 Stephenson
6523087 February 18, 2003 Busser
6535417 March 18, 2003 Tsuda et al.
6643748 November 4, 2003 Wieland
6725392 April 20, 2004 Frey et al.
6763455 July 13, 2004 Hall
6836816 December 28, 2004 Kendall
6985995 January 10, 2006 Holland et al.
7032125 April 18, 2006 Holt et al.
7047358 May 16, 2006 Lee et al.
7051155 May 23, 2006 Talagala et al.
7055058 May 30, 2006 Lee et al.
7065617 June 20, 2006 Wang
7069383 June 27, 2006 Yamamoto et al.
7076606 July 11, 2006 Orsley
7107480 September 12, 2006 Moshayedi et al.
7159150 January 2, 2007 Kenchammana-Hosekote et al.
7162575 January 9, 2007 Dalal et al.
7164608 January 16, 2007 Lee
7188270 March 6, 2007 Nanda et al.
7334156 February 19, 2008 Land et al.
7370220 May 6, 2008 Nguyen et al.
7386666 June 10, 2008 Beauchamp et al.
7398285 July 8, 2008 Kisley
7424498 September 9, 2008 Patterson
7424592 September 9, 2008 Karr et al.
7444532 October 28, 2008 Masuyama et al.
7480658 January 20, 2009 Heinla et al.
7484056 January 27, 2009 Madnani et al.
7484057 January 27, 2009 Madnani et al.
7484059 January 27, 2009 Ofer et al.
7536506 May 19, 2009 Ashmore et al.
7558859 July 7, 2009 Kasiolas et al.
7565446 July 21, 2009 Talagala et al.
7613947 November 3, 2009 Coatney et al.
7634617 December 15, 2009 Misra
7634618 December 15, 2009 Misra
7681104 March 16, 2010 Sim-Tang et al.
7681105 March 16, 2010 Sim-Tang et al.
7681109 March 16, 2010 Litsyn et al.
7730257 June 1, 2010 Franklin
7730258 June 1, 2010 Smith et al.
7730274 June 1, 2010 Usgaonkar
7743276 June 22, 2010 Jacobson et al.
7752489 July 6, 2010 Deenadhayalan et al.
7757038 July 13, 2010 Kitahara
7757059 July 13, 2010 Ofer et al.
7778960 August 17, 2010 Chatterjee et al.
7783955 August 24, 2010 Murin
7814272 October 12, 2010 Barrall et al.
7814273 October 12, 2010 Barrall
7818531 October 19, 2010 Barrall
7827351 November 2, 2010 Suetsugu et al.
7827439 November 2, 2010 Mathew et al.
7831768 November 9, 2010 Ananthamurthy et al.
7856583 December 21, 2010 Smith
7870105 January 11, 2011 Arakawa et al.
7873878 January 18, 2011 Belluomini et al.
7885938 February 8, 2011 Greene et al.
7886111 February 8, 2011 Klemm et al.
7908448 March 15, 2011 Chatterjee et al.
7916538 March 29, 2011 Jeon et al.
7921268 April 5, 2011 Jakob
7930499 April 19, 2011 Duchesne
7941697 May 10, 2011 Mathew et al.
7958303 June 7, 2011 Shuster
7971129 June 28, 2011 Watson et al.
7975115 July 5, 2011 Wayda et al.
7984016 July 19, 2011 Kisley
7991822 August 2, 2011 Bish et al.
8006126 August 23, 2011 Deenadhayalan et al.
8010485 August 30, 2011 Chatterjee et al.
8010829 August 30, 2011 Chatterjee et al.
8020047 September 13, 2011 Courtney
8046548 October 25, 2011 Chatterjee et al.
8051361 November 1, 2011 Sim-Tang et al.
8051362 November 1, 2011 Li et al.
8074038 December 6, 2011 Lionetti et al.
8082393 December 20, 2011 Galloway et al.
8086603 December 27, 2011 Nasre et al.
8086634 December 27, 2011 Mimatsu
8086911 December 27, 2011 Taylor
8090837 January 3, 2012 Shin et al.
8108502 January 31, 2012 Tabbara et al.
8117388 February 14, 2012 Jernigan, IV
8117521 February 14, 2012 Parker et al.
8140821 March 20, 2012 Raizen et al.
8145838 March 27, 2012 Miller et al.
8145840 March 27, 2012 Koul et al.
8175012 May 8, 2012 Chu et al.
8176360 May 8, 2012 Frost et al.
8176405 May 8, 2012 Hafner et al.
8180855 May 15, 2012 Aiello et al.
8200922 June 12, 2012 McKean et al.
8209469 June 26, 2012 Carpenter et al.
8225006 July 17, 2012 Karamcheti
8239618 August 7, 2012 Kotzur et al.
8244999 August 14, 2012 Chatterjee et al.
8261016 September 4, 2012 Goel
8271455 September 18, 2012 Kesselman
8285686 October 9, 2012 Kesselman
8305811 November 6, 2012 Jeon
8315999 November 20, 2012 Chatley et al.
8327080 December 4, 2012 Der
8335769 December 18, 2012 Kesselman
8341118 December 25, 2012 Drobychev et al.
8351290 January 8, 2013 Huang et al.
8364920 January 29, 2013 Parkison et al.
8365041 January 29, 2013 Olbrich et al.
8375146 February 12, 2013 Sinclair
8397016 March 12, 2013 Talagala et al.
8402152 March 19, 2013 Duran
8412880 April 2, 2013 Leibowitz et al.
8423739 April 16, 2013 Ash et al.
8429436 April 23, 2013 Fillingim et al.
8452928 May 28, 2013 Ofer et al.
8473698 June 25, 2013 Lionetti et al.
8473778 June 25, 2013 Simitci et al.
8473815 June 25, 2013 Chung et al.
8479037 July 2, 2013 Chatterjee et al.
8484414 July 9, 2013 Sugimoto et al.
8498967 July 30, 2013 Chatterjee et al.
8504797 August 6, 2013 Mimatsu
8522073 August 27, 2013 Cohen
8533408 September 10, 2013 Madnani et al.
8533527 September 10, 2013 Daikokuya et al.
8539177 September 17, 2013 Madnani et al.
8544029 September 24, 2013 Bakke et al.
8549224 October 1, 2013 Zeryck et al.
8583861 November 12, 2013 Ofer et al.
8589625 November 19, 2013 Colgrove et al.
8595455 November 26, 2013 Chatterjee et al.
8615599 December 24, 2013 Takefman et al.
8627136 January 7, 2014 Shankar et al.
8627138 January 7, 2014 Clark et al.
8639669 January 28, 2014 Douglis et al.
8639863 January 28, 2014 Kanapathippillai et al.
8640000 January 28, 2014 Cypher
8650343 February 11, 2014 Kanapathippillai et al.
8660131 February 25, 2014 Vermunt et al.
8661218 February 25, 2014 Piszczek et al.
8671072 March 11, 2014 Shah et al.
8689042 April 1, 2014 Kanapathippillai et al.
8700875 April 15, 2014 Barron et al.
8706694 April 22, 2014 Chatterjee et al.
8706914 April 22, 2014 Duchesneau
8706932 April 22, 2014 Kanapathippillai et al.
8712963 April 29, 2014 Douglis et al.
8713405 April 29, 2014 Healey, Jr. et al.
8719621 May 6, 2014 Karmarkar
8725730 May 13, 2014 Keeton et al.
8751859 June 10, 2014 Becker-Szendy et al.
8756387 June 17, 2014 Frost et al.
8762793 June 24, 2014 Grube et al.
8769232 July 1, 2014 Suryabudi et al.
8775858 July 8, 2014 Gower et al.
8775868 July 8, 2014 Colgrove et al.
8788913 July 22, 2014 Xin et al.
8793447 July 29, 2014 Usgaonkar et al.
8799746 August 5, 2014 Baker et al.
8819311 August 26, 2014 Liao
8819383 August 26, 2014 Jobanputra et al.
8822155 September 2, 2014 Sukumar et al.
8824261 September 2, 2014 Miller et al.
8832528 September 9, 2014 Thatcher et al.
8838541 September 16, 2014 Camble et al.
8838892 September 16, 2014 Li
8843700 September 23, 2014 Salessi et al.
8850108 September 30, 2014 Hayes et al.
8850288 September 30, 2014 Lazier et al.
8856593 October 7, 2014 Eckhardt et al.
8856619 October 7, 2014 Cypher
8862617 October 14, 2014 Kesselman
8862847 October 14, 2014 Feng et al.
8862928 October 14, 2014 Xavier et al.
8868825 October 21, 2014 Hayes et al.
8874836 October 28, 2014 Hayes et al.
8880793 November 4, 2014 Nagineni
8880825 November 4, 2014 Lionetti et al.
8886778 November 11, 2014 Nedved et al.
8898383 November 25, 2014 Yamamoto et al.
8898388 November 25, 2014 Kimmel
8904231 December 2, 2014 Coatney et al.
8918478 December 23, 2014 Ozzie et al.
8930307 January 6, 2015 Colgrove et al.
8930633 January 6, 2015 Amit et al.
8943357 January 27, 2015 Atzmony
8949502 February 3, 2015 McKnight et al.
8959110 February 17, 2015 Smith et al.
8959388 February 17, 2015 Kuang et al.
8972478 March 3, 2015 Storer et al.
8972779 March 3, 2015 Lee et al.
8977597 March 10, 2015 Ganesh et al.
8996828 March 31, 2015 Kalos et al.
9003144 April 7, 2015 Hayes et al.
9009724 April 14, 2015 Gold et al.
9021053 April 28, 2015 Bembo et al.
9021215 April 28, 2015 Meir et al.
9025393 May 5, 2015 Wu et al.
9043372 May 26, 2015 Makkar et al.
9047214 June 2, 2015 Northcott
9053808 June 9, 2015 Sprouse et al.
9058155 June 16, 2015 Cepulis et al.
9063895 June 23, 2015 Madnani et al.
9063896 June 23, 2015 Madnani et al.
9098211 August 4, 2015 Madnani et al.
9110898 August 18, 2015 Chamness et al.
9110964 August 18, 2015 Shilane et al.
9116819 August 25, 2015 Cope et al.
9117536 August 25, 2015 Yoon et al.
9122401 September 1, 2015 Zaltsman et al.
9123422 September 1, 2015 Yu et al.
9124300 September 1, 2015 Sharon et al.
9134908 September 15, 2015 Horn et al.
9153337 October 6, 2015 Sutardja
9158472 October 13, 2015 Kesselman et al.
9159422 October 13, 2015 Lee et al.
9164891 October 20, 2015 Karamcheti et al.
9183136 November 10, 2015 Kawamura et al.
9189650 November 17, 2015 Jaye et al.
9201733 December 1, 2015 Verma et al.
9207876 December 8, 2015 Shu et al.
9229656 January 5, 2016 Contreras et al.
9229810 January 5, 2016 He et al.
9235475 January 12, 2016 Shilane et al.
9244626 January 26, 2016 Shah et al.
9250999 February 2, 2016 Barroso
9251066 February 2, 2016 Colgrove et al.
9268648 February 23, 2016 Barash et al.
9268806 February 23, 2016 Kesselman
9280678 March 8, 2016 Redberg
9286002 March 15, 2016 Karamcheti et al.
9292214 March 22, 2016 Kalos et al.
9298760 March 29, 2016 Li et al.
9304908 April 5, 2016 Karamcheti et al.
9311969 April 12, 2016 Sharon et al.
9311970 April 12, 2016 Sharon et al.
9323663 April 26, 2016 Karamcheti et al.
9323667 April 26, 2016 Bennett
9323681 April 26, 2016 Apostolides et al.
9335942 May 10, 2016 Kumar et al.
9348538 May 24, 2016 Mallaiah et al.
9355022 May 31, 2016 Ravimohan et al.
9384082 July 5, 2016 Lee et al.
9384252 July 5, 2016 Akirav et al.
9389958 July 12, 2016 Sundaram et al.
9390019 July 12, 2016 Patterson et al.
9395922 July 19, 2016 Nishikido et al.
9396202 July 19, 2016 Drobychev et al.
9400828 July 26, 2016 Kesselman et al.
9405478 August 2, 2016 Koseki et al.
9411685 August 9, 2016 Lee
9417960 August 16, 2016 Cai et al.
9417963 August 16, 2016 He et al.
9430250 August 30, 2016 Hamid et al.
9430542 August 30, 2016 Akirav et al.
9432541 August 30, 2016 Ishida
9454434 September 27, 2016 Sundaram et al.
9471579 October 18, 2016 Natanzon
9477554 October 25, 2016 Hayes et al.
9477632 October 25, 2016 Du
9501398 November 22, 2016 George et al.
9525737 December 20, 2016 Friedman
9529542 December 27, 2016 Friedman et al.
9535631 January 3, 2017 Fu et al.
9552248 January 24, 2017 Miller et al.
9552291 January 24, 2017 Munetoh et al.
9552299 January 24, 2017 Stalzer
9563517 February 7, 2017 Natanzon et al.
9588698 March 7, 2017 Karamcheti et al.
9588712 March 7, 2017 Kalos et al.
9594652 March 14, 2017 Sathiamoorthy et al.
9600193 March 21, 2017 Ahrens et al.
9619321 April 11, 2017 Haratsch et al.
9619430 April 11, 2017 Kannan et al.
9645754 May 9, 2017 Li et al.
9667720 May 30, 2017 Bent et al.
9710535 July 18, 2017 Aizman et al.
9733840 August 15, 2017 Karamcheti et al.
9734225 August 15, 2017 Akirav et al.
9740403 August 22, 2017 Storer et al.
9740700 August 22, 2017 Chopra et al.
9740762 August 22, 2017 Horowitz et al.
9747319 August 29, 2017 Bestler et al.
9747320 August 29, 2017 Kesselman
9767130 September 19, 2017 Bestler et al.
9781227 October 3, 2017 Friedman et al.
9785498 October 10, 2017 Misra et al.
9798486 October 24, 2017 Singh
9804925 October 31, 2017 Carmi et al.
9811285 November 7, 2017 Karamcheti et al.
9811546 November 7, 2017 Bent et al.
9818478 November 14, 2017 Chung
9829066 November 28, 2017 Thomas et al.
9836245 December 5, 2017 Hayes et al.
9891854 February 13, 2018 Munetoh et al.
9891860 February 13, 2018 Delgado et al.
9892005 February 13, 2018 Kedem et al.
9892186 February 13, 2018 Akirav et al.
9904589 February 27, 2018 Donlan et al.
9904717 February 27, 2018 Anglin et al.
9910748 March 6, 2018 Pan
9910904 March 6, 2018 Anglin et al.
9934237 April 3, 2018 Shilane et al.
9940065 April 10, 2018 Kalos et al.
9946604 April 17, 2018 Glass
9952809 April 24, 2018 Shah
9959167 May 1, 2018 Donlan et al.
9965539 May 8, 2018 D'Halluin et al.
9998539 June 12, 2018 Brock et al.
10007457 June 26, 2018 Hayes et al.
10013177 July 3, 2018 Liu et al.
10013311 July 3, 2018 Sundaram et al.
10019314 July 10, 2018 Yang et al.
10019317 July 10, 2018 Usvyatsky et al.
10031703 July 24, 2018 Natanzon et al.
10061512 August 28, 2018 Lin
10073626 September 11, 2018 Karamcheti et al.
10082985 September 25, 2018 Hayes et al.
10089012 October 2, 2018 Chen et al.
10089174 October 2, 2018 Yang
10089176 October 2, 2018 Donlan et al.
10108819 October 23, 2018 Donlan et al.
10146787 December 4, 2018 Bashyam et al.
10152268 December 11, 2018 Chakraborty et al.
10157098 December 18, 2018 Yang et al.
10162704 December 25, 2018 Kirschner et al.
10180875 January 15, 2019 Klein
10185730 January 22, 2019 Bestler et al.
10235065 March 19, 2019 Miller et al.
10324639 June 18, 2019 Seo
10481798 November 19, 2019 Doshi et al.
10567406 February 18, 2020 Astigarraga et al.
10846137 November 24, 2020 Vallala et al.
10877683 December 29, 2020 Wu et al.
11076509 July 27, 2021 Alissa et al.
11106810 August 31, 2021 Natanzon et al.
11119657 September 14, 2021 Doshi et al.
11194707 December 7, 2021 Stalzer
20020144059 October 3, 2002 Kendall
20030105984 June 5, 2003 Masuyama et al.
20030110205 June 12, 2003 Johnson
20040161086 August 19, 2004 Buntin et al.
20050001652 January 6, 2005 Malik et al.
20050076228 April 7, 2005 Davis et al.
20050235132 October 20, 2005 Karr et al.
20050278460 December 15, 2005 Shin et al.
20050283649 December 22, 2005 Turner et al.
20060015683 January 19, 2006 Ashmore et al.
20060114930 June 1, 2006 Lucas et al.
20060174157 August 3, 2006 Barrall et al.
20060248294 November 2, 2006 Nedved et al.
20070079068 April 5, 2007 Draggon
20070214194 September 13, 2007 Reuter
20070214314 September 13, 2007 Reuter
20070234016 October 4, 2007 Davis et al.
20070268905 November 22, 2007 Baker et al.
20080080709 April 3, 2008 Michtchenko et al.
20080107274 May 8, 2008 Worthy
20080155191 June 26, 2008 Anderson et al.
20080256141 October 16, 2008 Wayda et al.
20080295118 November 27, 2008 Liao
20090077208 March 19, 2009 Nguyen et al.
20090138654 May 28, 2009 Sutardja
20090216910 August 27, 2009 Duchesneau
20090216920 August 27, 2009 Lauterbach et al.
20100017444 January 21, 2010 Chatterjee et al.
20100042636 February 18, 2010 Lu
20100094806 April 15, 2010 Apostolides et al.
20100115070 May 6, 2010 Missimilly
20100125695 May 20, 2010 Wu et al.
20100162076 June 24, 2010 Sim-Tang et al.
20100169707 July 1, 2010 Mathew et al.
20100174576 July 8, 2010 Naylor
20100223423 September 2, 2010 Sinclair
20100268908 October 21, 2010 Ouyang et al.
20100306500 December 2, 2010 Mimatsu
20110035540 February 10, 2011 Fitzgerald et al.
20110040925 February 17, 2011 Frost et al.
20110060927 March 10, 2011 Fillingim et al.
20110119462 May 19, 2011 Leach et al.
20110219170 September 8, 2011 Frost et al.
20110238625 September 29, 2011 Hamaguchi et al.
20110264843 October 27, 2011 Haines et al.
20110302369 December 8, 2011 Goto et al.
20120011398 January 12, 2012 Eckhardt et al.
20120079318 March 29, 2012 Colgrove et al.
20120089567 April 12, 2012 Takahashi et al.
20120110249 May 3, 2012 Jeong et al.
20120131253 May 24, 2012 McKnight et al.
20120158923 June 21, 2012 Mohamed et al.
20120191900 July 26, 2012 Kunimatsu et al.
20120198152 August 2, 2012 Terry et al.
20120198261 August 2, 2012 Brown et al.
20120209943 August 16, 2012 Jung
20120226934 September 6, 2012 Rao
20120246435 September 27, 2012 Meir et al.
20120260055 October 11, 2012 Murase
20120311557 December 6, 2012 Resch
20130022201 January 24, 2013 Glew et al.
20130036314 February 7, 2013 Glew et al.
20130042056 February 14, 2013 Shats et al.
20130060884 March 7, 2013 Bernbo et al.
20130067188 March 14, 2013 Mehra et al.
20130073894 March 21, 2013 Xavier et al.
20130124776 May 16, 2013 Hallak et al.
20130132800 May 23, 2013 Healey, Jr. et al.
20130151653 June 13, 2013 Sawicki et al.
20130151771 June 13, 2013 Tsukahara et al.
20130173853 July 4, 2013 Ungureanu et al.
20130238554 September 12, 2013 Yucel et al.
20130339314 December 19, 2013 Carpentier et al.
20130339635 December 19, 2013 Amit et al.
20130339818 December 19, 2013 Baker et al.
20140040535 February 6, 2014 Lee et al.
20140040702 February 6, 2014 He et al.
20140047263 February 13, 2014 Coatney et al.
20140047269 February 13, 2014 Kim
20140063721 March 6, 2014 Herman et al.
20140064048 March 6, 2014 Cohen et al.
20140068224 March 6, 2014 Fan et al.
20140075252 March 13, 2014 Luo et al.
20140122510 May 1, 2014 Namkoong et al.
20140136880 May 15, 2014 Shankar et al.
20140181402 June 26, 2014 White
20140220561 August 7, 2014 Sukumar et al.
20140237164 August 21, 2014 Le et al.
20140279936 September 18, 2014 Bernbo et al.
20140280025 September 18, 2014 Eidson et al.
20140289588 September 25, 2014 Nagadomi et al.
20140330785 November 6, 2014 Isherwood et al.
20140372838 December 18, 2014 Lou et al.
20140380125 December 25, 2014 Calder et al.
20140380126 December 25, 2014 Yekhanin et al.
20150032720 January 29, 2015 James
20150039645 February 5, 2015 Lewis
20150039849 February 5, 2015 Lewis
20150089283 March 26, 2015 Kermarrec et al.
20150100746 April 9, 2015 Rychlik et al.
20150134824 May 14, 2015 Mickens et al.
20150153800 June 4, 2015 Lucas et al.
20150154418 June 4, 2015 Redberg
20150180714 June 25, 2015 Chunn et al.
20150280959 October 1, 2015 Vincent
20160026397 January 28, 2016 Nishikido et al.
20160182542 June 23, 2016 Staniford
20160191508 June 30, 2016 Bestler et al.
20160246537 August 25, 2016 Kim
20160248631 August 25, 2016 Duchesneau
20160378612 December 29, 2016 Hipsh et al.
20170091236 March 30, 2017 Hayes et al.
20170103092 April 13, 2017 Hu et al.
20170103094 April 13, 2017 Hu et al.
20170103098 April 13, 2017 Hu et al.
20170103116 April 13, 2017 Hu et al.
20170177236 June 22, 2017 Haratsch et al.
20170262202 September 14, 2017 Seo
20180039442 February 8, 2018 Shadrin et al.
20180054454 February 22, 2018 Astigarraga et al.
20180081958 March 22, 2018 Akirav et al.
20180101441 April 12, 2018 Hyun et al.
20180101587 April 12, 2018 Anglin et al.
20180101588 April 12, 2018 Anglin et al.
20180217756 August 2, 2018 Liu et al.
20180307560 October 25, 2018 Vishnumolakala et al.
20180321874 November 8, 2018 Li et al.
20190036703 January 31, 2019 Bestler
20190220315 July 18, 2019 Vallala et al.
20200034560 January 30, 2020 Natanzon et al.
20200326871 October 15, 2020 Wu et al.
20210360833 November 18, 2021 Alissa et al.
Foreign Patent Documents
2164006 March 2010 EP
2256621 December 2010 EP
0213033 February 2002 WO
2008103569 August 2008 WO
2008157081 December 2008 WO
2013032825 March 2013 WO
Other references
  • Hwang et al., “RAID-x: A New Distributed Disk Array for I/O-Centric Cluster Computing”, Proceedings of The Ninth International Symposium on High-performance Distributed Computing, Aug. 2000, pp. 279-286, The Ninth International Symposium on High-Performance Distributed Computing, IEEE Computer Society, Los Alamitos, CA.
  • International Search Report and Written Opinion, PCT/US2015/018169, dated May 15, 2015, 10 pages.
  • International Search Report and Written Opinion, PCT/US2015/034291, dated Sep. 30, 2015, 3 pages.
  • International Search Report and Written Opinion, PCT/US2015/034302, dated Sep. 11, 2015, 10 pages.
  • International Search Report and Written Opinion, PCT/US2015/039135, dated Sep. 18, 2015, 8 pages.
  • International Search Report and Written Opinion, PCT/US2015/039136, dated Sep. 23, 2015, 7 pages.
  • International Search Report and Written Opinion, PCT/US2015/039137, dated Oct. 1, 2015, 8 pages.
  • International Search Report and Written Opinion, PCT/US2015/039142, dated Sep. 24, 2015, 3 pages.
  • International Search Report and Written Opinion, PCT/US2015/044370, dated Dec. 15, 2015, 3 pages.
  • International Search Report and Written Opinion, PCT/US2016/014356, dated Jun. 28, 2016, 3 pages.
  • International Search Report and Written Opinion, PCT/US2016/014357, dated Jun. 29, 2016, 3 pages.
  • International Search Report and Written Opinion, PCT/US2016/014361, dated May 30, 2016, 3 pages.
  • International Search Report and Written Opinion, PCT/US2016/014604, dated May 19, 2016, 3 pages.
  • International Search Report and Written Opinion, PCT/US2016/016504, dated Jul. 6, 2016, 7 pages.
  • International Search Report and Written Opinion, PCT/US2016/023485, dated Jul. 21, 2016, 13 pages.
  • International Search Report and Written Opinion, PCT/US2016/024391, dated Jul. 12, 2016, 11 pages.
  • International Search Report and Written Opinion, PCT/US2016/026529, dated Jul. 19, 2016, 9 pages.
  • International Search Report and Written Opinion, PCT/US2016/031039, dated Aug. 18, 2016, 7 pages.
  • International Search Report and Written Opinion, PCT/US2016/033306, dated Aug. 19, 2016, 11 pages.
  • International Search Report and Written Opinion, PCT/US2016/047808, dated Nov. 25, 2016, 14 pages.
  • Kim et al., “Data Access Frequency based Data Replication Method using Erasure Codes in Cloud Storage System”, Journal of the Institute of Electronics and Information Engineers, Feb. 2014, vol. 51, No. 2, 7 pages.
  • Schmid, “RAID Scaling Charts, Part 3:4-128 KB Stripes Compared”, Tom's Hardware, Nov. 27, 2007, URL: http://www.tomshardware.com/reviews/RAID-SCALING-CHARTS.1735-4.html, 24 pages.
  • Stalzer, “FlashBlades: System Architecture and Applications”, Proceedings of the 2nd Workshop on Architectures and Systems for Big Data, Jun. 2012, pp. 10-14, Association for Computing Machinery, New York, NY.
  • Storer et al., “Pergamum: Replacing Tape with Energy Efficient, Reliable, Disk-Based Archival Storage”, FAST'08: Proceedings of the 6th USENIX Conference on File and Storage Technologies, Article No. 1, Feb. 2008, pp. 1-16, USENIX Association, Berkeley, CA.
Patent History
Patent number: 11995318
Type: Grant
Filed: Apr 6, 2023
Date of Patent: May 28, 2024
Patent Publication Number: 20230244382
Assignee: PURE STORAGE, INC. (Santa Clara, CA)
Inventors: Eric D. Seppanen (Mountain View, CA), Neil Buda Vachharajani (Menlo Park, CA), Nidhi Pankaj Doshi (Menlo Park, CA)
Primary Examiner: Mardochee Chery
Application Number: 18/296,880
Classifications
Current U.S. Class: Accessing, Addressing Or Allocating Within Memory Systems Or Architectures (epo) (711/E12.001)
International Classification: G06F 12/00 (20060101); G06F 3/06 (20060101); G06F 12/02 (20060101);